Abstract While functional magnetic resonance (fMRI) has proved invaluable for identifying where in the brain activation is occurring during a particular task, it has had less to say about how the dynamics of that activation actually contribute to task performance. Indeed, because of the belief that fMRI signals are sluggish and temporally imprecise, fMRI experimental paradigms traditionally have used sustained block designs which deliberately preclude measuring the rapid changes in sensory and motor signals that underlie everyday actions. Recent evidence, however, suggests that there is considerable temporal information present in the BOLD signal, opening the possibility that fast neuronal dynamics can be revealed by fMRI. In this proposal, we will examine this possibility with a series of multimodal experiments in which a consistent experimental paradigm is applied across spatial and temporal scales to quantify responses to transient inputs. In the first specific aim, we will characterize the relationship between cellular level population activity,mesoscopic activity, and metabolic patterns across variations in input strength, network state, and behavioral state using simultaneous 2-photon and 1-photon imaging in ferret visual cortex. In the second specific aim, we will extend these results to awake behaving primates using 1-photon imaging to resolve and compare metabolic and hemodynamic responses. In the third specific aim, we will extend these results to fast whole brain fMRI in both monkeys and humans. We will integrate the results aims 1 and 2, in which forward spatiotemporal filters describing the evolution of impulses of activity to metabolism and consequently to hemodynamics are measured, to construct an inverse filter that allows temporally precise inferences regarding neuronal activity from fast fMRI measurements. The same manipulations of local and network inputs used in the aims 1 and 2 will also be used to relate the variability of local event-driven signals to more global fluctuations such as are seen in resting state studies. These innovations will transform fMRI into a non-invasive systems neuroscience paradigm that can reveal not only spatial aspects of neural activity but also measure the temporal dynamics of neural activity that underlie cognitive processes and behavior.